Merge pull request #3827 from pipecat-ai/pk/gemini-tts-service-remove-model-ivar

Remove unnecessary `_model` ivar from `GeminiTTSService`, using `_set…

src/pipecat/services/elevenlabs/stt.py

@@ -279,7 +279,6 @@ class ElevenLabsSTTService(SegmentedSTTService):
         self._api_key = api_key
         self._base_url = base_url
         self._session = aiohttp_session
-        self._model_id = model
 
     def can_generate_metrics(self) -> bool:
         """Check if the service can generate processing metrics.
@@ -300,25 +299,6 @@ class ElevenLabsSTTService(SegmentedSTTService):
         """
         return language_to_elevenlabs_language(language)
 
-    async def _update_settings(self, delta: STTSettings) -> dict[str, Any]:
-        """Apply a settings delta.
-
-        Converts language to ElevenLabs format before applying and keeps
-        ``_model_id`` in sync with the model setting.
-
-        Args:
-            delta: A :class:`STTSettings` (or ``ElevenLabsSTTSettings``) delta.
-
-        Returns:
-            Dict mapping changed field names to their previous values.
-        """
-        changed = await super()._update_settings(delta)
-
-        if "model" in changed:
-            self._model_id = self._settings.model
-
-        return changed
-
     async def _transcribe_audio(self, audio_data: bytes) -> dict:
         """Upload audio data to ElevenLabs and get transcription result.
 
@@ -344,7 +324,7 @@ class ElevenLabsSTTService(SegmentedSTTService):
         )
 
         # Add required model_id, language_code, and tag_audio_events
-        data.add_field("model_id", self._model_id)
+        data.add_field("model_id", self._settings.model)
         data.add_field("language_code", self._settings.language)
         data.add_field("tag_audio_events", str(self._settings.tag_audio_events).lower())
 
@@ -522,7 +502,6 @@ class ElevenLabsRealtimeSTTService(WebsocketSTTService):
 
         self._api_key = api_key
         self._base_url = base_url
-        self._model_id = model
         self._audio_format = ""  # initialized in start()
         self._receive_task = None
 
@@ -540,9 +519,6 @@ class ElevenLabsRealtimeSTTService(WebsocketSTTService):
     async def _update_settings(self, delta: STTSettings) -> dict[str, Any]:
         """Apply a settings delta and reconnect if anything changed.
 
-        Converts language to ElevenLabs format before applying and keeps
-        ``_model_id`` in sync.
-
         Args:
             delta: A :class:`STTSettings` (or ``ElevenLabsRealtimeSTTSettings``) delta.
 
@@ -554,11 +530,9 @@ class ElevenLabsRealtimeSTTService(WebsocketSTTService):
         if not changed:
             return changed
 
-        if "model" in changed:
-            self._model_id = self._settings.model
-
         await self._disconnect()
         await self._connect()
+
         return changed
 
     async def start(self, frame: StartFrame):
@@ -704,7 +678,7 @@ class ElevenLabsRealtimeSTTService(WebsocketSTTService):
             logger.debug("Connecting to ElevenLabs Realtime STT")
 
             # Build query parameters
-            params = [f"model_id={self._model_id}"]
+            params = [f"model_id={self._settings.model}"]
 
             if self._settings.language:
                 params.append(f"language_code={self._settings.language}")

src/pipecat/services/google/tts.py

@@ -1236,7 +1236,7 @@ class GeminiTTSService(GoogleBaseTTSService):
         super().__init__(
             sample_rate=sample_rate,
             settings=GeminiTTSSettings(
-                model=None,
+                model=model,
                 language=self.language_to_service_language(params.language)
                 if params.language
                 else "en-US",
@@ -1249,7 +1249,6 @@ class GeminiTTSService(GoogleBaseTTSService):
         )
 
         self._location = location
-        self._model = model
         self._client: texttospeech_v1.TextToSpeechAsyncClient = self._create_client(
             credentials, credentials_path
         )
@@ -1327,7 +1326,7 @@ class GeminiTTSService(GoogleBaseTTSService):
 
                 voice = texttospeech_v1.VoiceSelectionParams(
                     language_code=self._settings.language,
-                    model_name=self._model,
+                    model_name=self._settings.model,
                     multi_speaker_voice_config=multi_speaker_voice_config,
                 )
             else:
@@ -1335,7 +1334,7 @@ class GeminiTTSService(GoogleBaseTTSService):
                 voice = texttospeech_v1.VoiceSelectionParams(
                     language_code=self._settings.language,
                     name=self._settings.voice,
-                    model_name=self._model,
+                    model_name=self._settings.model,
                 )
 
             # Create streaming config

src/pipecat/services/hathora/stt.py

@@ -101,8 +101,6 @@ class HathoraSTTService(SegmentedSTTService):
             ),
             **kwargs,
         )
-
-        self._model = model
         self._api_key = api_key or os.getenv("HATHORA_API_KEY")
         self._base_url = base_url
 
@@ -136,7 +134,7 @@ class HathoraSTTService(SegmentedSTTService):
             url = f"{self._base_url}"
 
             payload = {
-                "model": self._model,
+                "model": self._settings.model,
             }
 
             if self._settings.language is not None:

src/pipecat/services/hathora/tts.py

@@ -120,7 +120,6 @@ class HathoraTTSService(TTSService):
             ),
             **kwargs,
         )
-        self._model = model
         self._api_key = api_key or os.getenv("HATHORA_API_KEY")
         self._base_url = base_url
 
@@ -149,7 +148,7 @@ class HathoraTTSService(TTSService):
 
             url = f"{self._base_url}"
 
-            payload = {"model": self._model, "text": text}
+            payload = {"model": self._settings.model, "text": text}
 
             if self._settings.voice is not None:
                 payload["voice"] = self._settings.voice